This is amazing. Now I'm imagining the ability to simulate less-than-ideal conditions for tree growth. Placing the tree next to a building or on a hill or culling branches at certain stages in its life to see what happens next, etc.
I think my overall thought is: we see plenty of nice looking trees in games and graphics. What about authentically degenerate trees?
Now I want a simulated caterpillar. In my area, we sometimes get caterpillar apocalypse where the whole forest will be full of them, sometimes they cover entire trees with their silk.
I think the most interesting result would be somewhere halfway between 0 and 100% nibbled.
Don't stop there: fungi and beetles are some other stressors that could be modeled alongside caterpillars using similar techniques. Like lepidopterans, they are very host-specific, though they have somewhat different transmission vectors, so you could see the defoliation (or other foliage impacts like leaf curling and yellowing) spread / not spread depending on factors such as proximity, moisture, wind and the parasite's lifecycle.
Pretty cool idea: "buds compete for light, the tree distributes a growth resource among them, and the successful ones produce new shoots. Those shoots change the light available for the next round. The crown shape emerges from this feedback loop."
I would never have thought to generate tree shapes by simulating real life
There has been a complete deluge of new accounts sometimes only hours or days old being extremely condescending, overly negative, sometimes antagonistic, and sometimes they seem like a kind of LLM prompted to cause arguments.
This is one of my favorite topics recently — a bit of a guilty pleasure. Please share other procedural flora growth if you know of examples!
I’ve found it’s a lot, lot harder than I expected, though I’m targeting aroids (monstera, philodendron, pothos, anthurium, etc) rather than trees, which adds the tricky requirement to get the leaves to unfurl naturally from their petioles. Not quite completely discouraged yet though, sadly!
The generated trees look absolutely amazing. Adjusting the process for different lighting situations seems straightforward too - freestanding trees look different from those in woods.
All their growth models do involve light, however. So it probably does make sense to have a common light representation - ideally some sort of sparse cone storage though, not a voxel grid the size of a forest.
I think my overall thought is: we see plenty of nice looking trees in games and graphics. What about authentically degenerate trees?
I think the most interesting result would be somewhere halfway between 0 and 100% nibbled.
I would never have thought to generate tree shapes by simulating real life
I’ve found it’s a lot, lot harder than I expected, though I’m targeting aroids (monstera, philodendron, pothos, anthurium, etc) rather than trees, which adds the tricky requirement to get the leaves to unfurl naturally from their petioles. Not quite completely discouraged yet though, sadly!